We present a new algorithm, called marching cubes, that creates triangle models of constant density surfaces from 3D medical data. Using a divide-and-conquer approach to generate inter-slice connectivity, we create a case table that defines triangle topology. The algorithm processes the 3D medical data in scan-line order and calculates triangle vertices using linear interpolation. We find the gradient of the original data, normalize it, and use it as a basis for shading the models. The detail in images produced from the generated surface models is the result of maintaining the inter-slice connectivity, surface data, and gradient information present in the original 3D data. Results from computed tomography (CT), magnetic resonance (MR), and single-photon emission computed tomography (SPECT) illustrate the quality and functionality of marching cubes. We also discuss improvements that decrease processing time and add solid modeling capabilities.

In this paper we evaluate the performance of a seed filling algorithm operating in view lattice when rendering maximum intensity projections (MIP). We evaluate the combination of the seed filling algorithm and the template algorithm. We show that the template algorithm is a particularly attractive companion to the seed filling algorithm, because the template algorithm allows stepping in all six directions within the resampled data set with very few instructions and without conditional branches. The results also indicate that given a reasonable low-level threshold for empty voxels the MIP projection can be calculated in real time with standard desktop workstations. In addition we observe that the performance of our seed filling algorithm stays relatively constant with different viewing angles, as opposed to the template algorithm alone. Key Words: Volume Visualization, Three-dimensional (3D) Imaging, Ray Casting, Volume Rendering, Medical Applications. INTRODUCTION In this ...

ix Preface x Acknowledgements : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : x Publication History : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : xi 1. Introduction 1 2. Scientific Visualization: Environments 5 2.1 Some Scientific Visualization Environments : : : : : : : : : : : : : : : : : : 5 2.1.1 Turnkey Visualization Systems : : : : : : : : : : : : : : : : : : : : : 6 2.1.2 Extensible Visualization Systems : : : : : : : : : : : : : : : : : : : : 7 2.2 Spray Rendering : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 13 2.3 How Mix&Match relates to other environments and spray rendering : : : : 14 3. Scientific Visualization: Techniques 17 3.1 Scientific Data : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 17 3.2 A Classification : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 18 3.3 Two Dimensional Visualization : : : : : : : : : : : : : : : : : : : : : : : : : 21 3.4 Volume Visua...

The dramatic increase in the use of 3D image acquisition devices over the past decade has inspired major new developments in the display of volume data sets. In this chapter we present an overview of these diverse display methods and discuss the relative advantages and disadvantages of each of the different approaches. In addition, we touch upon some of the major issues involved in creating high-quality images from volume data, including the problems of surface definition and object segmentation. Due in part to the rapid, almost frantic, pace of recent developments in methods for rendering images from volume data, there has not yet emerged any widely accepted taxonomy for these methods. Because the human visual system is adapted for environments in which images of surfaces predominate, most algorithms emphasize in one way or another the display of surface-like information, either implicitly or explicitly. For clarity, we will avoid using the terms "surface rendering " and "volume rendering " to describe the various methods, since although prevalent in the literature they have no precise, commonly accepted definitions. Instead, we will differentiate the various rendering methods using the following three characteristics, which are somewhat more precise and, we hope, less misleading: 1) whether the explicit creation of an intermediate surface representation is required (if so,